The new Twitter API constraints make it very expensive ($42000) / slow (months) to download this dataset. Can it be distributed another way?